Leveraging AI

113 | OpenAI leadership Exodus continues, AI cost competition intensifies, Google search dominance at risk? And other important AI news for the week ending on August 9 2024

• Isar Meitis • Episode 113

Share episode

What happens when the pioneers of AI start abandoning ship?

In this episode of Leveraging AI, we dive into a whirlwind of developments shaking up the AI landscape—from the unexpected departure of three key figures at OpenAI to the drastic price cuts in advanced AI models. Is this the beginning of the end for OpenAI’s dominance, or just the next phase of its evolution?

We’ll talk about what’s driving this brain drain, including the potential fallout of these high-profile exits on AI safety and business strategies. With fierce competition from Anthropic and other open-source models, the stakes are higher than ever.

So, what does this mean for your business? As AI becomes more accessible and affordable, companies must adapt quickly or risk being left behind. We’ll explore the implications of these shifts and how to stay ahead in the rapidly evolving AI market.

In this episode, you’ll discover:
- The reasons behind the departure of OpenAI’s co-founders and top executives.
- How price slashing in AI models could affect your business strategy.
- The rising competition between closed-source giants and open-source challengers.
- What the future holds for AI safety and alignment with business goals.


About Leveraging AI

If you’ve enjoyed or benefited from some of the insights of this episode, leave us a five-star review on your favorite podcast platform, and let us know what you learned, found helpful, or liked most about this show!

Hello and welcome to a weekend news edition of the Leveraging AI podcast, the podcast that shares practical ethical ways to improve efficiency, grow your business and advance your career. This is Issar Mehtis, your host, and we have a jam packed week of news today, including a significant decline in the cost of using advanced AI models, release of a few new open source models from some interesting sources, and the departure of three key figures from open AI. Two of them are founders. So lots to talk about. So let's get started. And we'll start with the big news from OpenAI. As I mentioned, three key figures left the company. So it seems that every time we thought we've seen Everything that we can, as far as the term oil from a personnel perspective in open AI, every time we think it's over, something new happens. And it's another big one. So two of the co founders, one of them is John Shulman, which has been with open AI from the beginning. One of the researchers behind a lot of the stuff and the systems and the capabilities that we know is living to Anthropic. So open AI is largest competitor from the closed source world, other than Google. And he's leaving to Anthropic saying that he wants to focus more on superalignment, which means on AI safety, which is the area that got maybe the biggest hit in open AI with Ilya Saskovor and Yanliki, which has left after running that division and that division more or less dissolved. So another hint on maybe the lack of focus on safety within open AI. So John Shulman leaving to do exactly that, the same thing that he claims that he's passionate about. Working for Anthropic. The other big departure is Greg Brockman. So Greg Brockman is one of the co founders and the president of OpenAI. And he's not leaving. He's taking an extended leave of absence, a sabbatical through the end of 2024. But I think that's a very big hint where the wind is blowing. I want to remind you, roll back to the Past year in senior departures and changes in open AI. So at the end of 2023, the board decided to axe Sam Altman and get him out of the company. He was let go of the company. And then in a big term oil that involved obviously Microsoft as a big investor that pulled a lot of strings, as well as employees within open AI, He was brought back as the CEO again. One of the people that supported him very strongly was Greg Brockman, who was the president before that and resigned from his position following the oust of Sam Altman. And he was brought back and reinstated as president when Sam got back as well. So his departure is very interesting from a political point of view. Politics perspective, because he was a very strong ally for Sam Altman. No information was released from OpenAI itself. The third big figure that left is Peter Deng. Now that's a less known name, but he's a very interesting position. So Peter was brought in in November of 2023 to lead product development in open AI. He has deep experience in running product leadership at meta Uber and air table. So he was brought in basically to help open AI commercialize and build better products, and he's leaving less than a year after coming over. It's also not a good sign for top leadership in the direction that OpenAI is taking. So out of the original 11 founders of OpenAI, only three are left in the company. One of them, and the most important one is obviously Sam Altman. Everybody else left many of them in this past year. Another big change that happened from a safety perspective in OpenAI recently that we talked about was Alexander Madry, that was leading the safety team, has also left. was reassigned to another role. So overall, lots of big senior changes in OpenAI. This is never good news when the core leadership of a team is either leaving or changing places and especially moving to the competition and a few of the people move to Anthropic, it's never a good sign. Now, I don't know what that means to OpenAI. We talked last week about their very, very significant. And deep losses that they're expected to experience this year. the assumption is about 5 billion in loss. It will be very interesting to see where that leads the company and how much that turn off term oil is going to impact its performance, but let's dive into some additional details about open AI. And that's going to show you where they are, maybe more both from a technical perspective, as well as from a business perspective. If you remember a couple of weeks ago, there was a rain of new models and new faster models and cheaper models. And I told you that one of the things that I'm very curious about is where does that lead from a business model perspective when models are getting better and better. And at the same time, cheaper and cheaper to operate a because smaller models can outperform very well and B because there's fierce competition and a lot of open source models are coming into the market that are highly capable. Where does that leave companies like open AI and Anthropic that are trying to run a closed sourced enterprise level solution for this universe of AI solutions. And I. Don't have a good answer. And I don't know if there is a good answer, but as of right now, there are serious signs that the competition is only getting stronger. So open AI without announcing it, just cut the costs of using GPT 4. 0 by 50%. So there's a new model that was just released. It's called GPT 4. 0 2024 8 06. That's basically states the dates when it was released. It is supposedly slightly better than the previous GPT 4. 0, and it's 50 percent cheaper to use through the API. This comes obviously from significant pressure by the new models by metalama 3. 1, which does very well. And it was cheaper than that. And Gemini 1. 5 Pro that is cheaper than that. And from Anthropic 3. 5 Sonnet, which is a very capable model. That is roughly priced the same way. So this new model comes to make open AI GPT 4. Oh, more competitive to developers to keep on using. They also cut the cost for fine tuning GPT 4. 0 mini. So I mentioned last week that they've now enabling companies to fine tune GPT 4. 0 mini again to compete with open source model that are becoming better and closer to its capabilities. They've cut the cost for that dramatically as well and increased the capabilities of the model So now the data for training can be up to 65, 000 tokens, which is significantly more than it was before. That said, the fine tuning functionality currently only exists for the text functionality and not for images. So they're dropping their prices dramatically. Now, in addition to that, there's some other interesting news that are coming from the OpenAI direction. As we all know, they've been training and working on GPT 5 for a very long time. Nobody's really telling us what's in GPT 5 or when exactly it's going to be released, other than a lot of rumors and hints from Sam Altman saying that the current models that they have is quote unquote is quote embarrassing at best. And so they're working on something much bigger, but they have been releasing more and more capabilities on the side. So the voice capability that has been rolling out to a few test users, and we're going to talk more about that in a minute, as well as demoing of video capabilities and so on. This could be our keyhole to look into the room of GPT 5 and starting to understand some of the capabilities that it's going to bring to us. So maybe that's part of the strategy, and that's what OpenAI has always done, is just releasing bits and pieces of things and iterating through its releases versus one big release of everything. I still think there's going to be a release of GPT 5, but we might be looking at some of these capabilities already through these iterative things. There's a rumored Project Strawberry that is brewing within OpenAI. Nobody knows exactly what it is. If you remember when we go back, there were a lot of rumors about Project Q Star that may have led to the axing of Sam Altman that I mentioned earlier back in 2023. So now there's this new secretive project called Project Strawberry And Sam Altman tweeted this week an image of strawberries and all he wrote was, I love summer in the garden, but with strawberries, everybody obviously jumped into conclusions that I has to do something with Project Strawberry that might be really soon. So we'll keep you posted learn what that is. In parallel, a new model appeared on LMC's chatbot arena, which is a platform we talked about many times before that allows users to rank models based on their performance at a blank test. So you basically give it a score running two side by side, speaking the one you think is working better. And this is how they rank AI models based on actual people's usage without people knowing what they're using. In my perspective, this is the most relevant leaderboard right now. So a new model called anonymous chatbot showed up there and it's actually doing pretty well. And the model itself, when asked what model it is, it's claiming it's based on GPT 4 architecture. OpenAI has done stuff like that before, basically released models not taking responsibility for them or claiming them as theirs, but testing them on the chatbot arena to see how they perform. So multiple things are brewing in the back end, despite everything that's going on the personnel level and the price wars with the competition, which we're going to talk more about in a minute. Now, two interesting things that have been in the back end that OpenAI has been talking about. One is they've been working on a invisible text watermarking for everything that they're generating. That's basically a way for them to allow other organizations to detect Any text that was generated by any chat GPT models. All it does is it makes slightly small changes to the pick of words from the algorithm that they can be detected by their own detection tool. So from one perspective, that's interesting because it will allow us to detect what chat GPT creates. But there apparently several different problems with that. One is it's not the only model. It can only detect what ChachiPT has created. Two is that they're afraid that bad actors can circumvent that change in the algorithm and hence still allowing the wrong people to create text that nobody knows was created with ChachiPT. And They have an issue with the fact that it's disproportionately impacting non English speakers. So people who speak English can write English very well, so their need to use a tool like Chachapiti to help them write in English is significantly smaller than English people who are non native English speakers, and so they're afraid from that impact as well. So that tool, while it exists right now, is not going to be released, at least not for now. Now, we talked about last week that OpenAI started rolling out Their voice functionality of GPT 4. 0 to remind everybody they demo that capability earlier this year, a day before Google's big announcement, and they showed a model that can have a human like conversation across the world. Multiple use cases, including having multiple kinds of conversations, and the biggest difference between that and the current voice capability is that it's actually multimodal, meaning the previous model would actually do three steps to respond to you. Step one, it would transcribe what you say. Step two, it will analyze and provide an answer just like a regular chat. And step three, it will translate the words in the regular chat into voice. So that took about five seconds on average to get a response. And this new model just talks like a regular human story, talking about milliseconds in the delay of its response. It literally just like you and I, you don't Transcribed Take steps when you're having a conversation with somebody, you just know what to say, and you say it, and this works the same way, including the capability to interrupt it in the middle of a sentence and have it continue with its logic. So very powerful capabilities, and they just started rolling it out, but they also just shared what they call a system card for the GPT 4. 0, stating some of the risks and why they're still working with it and haven't released it yet. I'll share a link to that card. It's pretty long, and it details a lot of reasons why they're not releasing it yet and what might be the issues, but the two very interesting things that they mentioned there. One is, That they're acknowledging that users may become emotionally attached to a human like voice interface. The researchers that evaluated multiple people using it has observed language that's indicating emotional connection during the testing. So that's a very short period amount of time. Don't mention the fact that this might be your day to day assistant and you may find yourself talking to it more than you talk to, other people, including your spouse and other employees in your companies and so on. Because that might become, and most likely will become, a problem. Our interface to computers. We won't use keyboard. Keyboard doesn't make any sense. It's not an efficient way of communication. And part of the fear is that the persuasive capabilities that chats already have over us will grow significantly with voice capabilities, especially that this voice capability can mimic and manipulate human emotion that may lead to a lot of bad things, including obviously hurting real human relationships and social interactions in general, as well as People will start building trust in these models and being dependent on them for multiple things, which obviously has a lot of other negative impacts. Another really weird, scary, and interesting thing that happened is that the AI started mimicking the voices of the people it was speaking with. So basically duplicating the user and using his voice. Nuances of the voice, the way he speaks, his diction, as well as the voice itself without being asked to do that. That's again, very scary on multiple levels and very interesting on why and how the AI is doing it, but it's not something they definitely want to release into the wild. Another thing that they're afraid of is all these models have various guardrails to keep us safe from other things that they might be able to do and we don't necessarily want people to do with them and they fear that the ability to have a conversation will enable new jailbreaking capabilities that will allow people and bad actors to put these models to a bad use. So overall, lots of good reasons why not release the model yet. But as I mentioned, they already started releasing it to a small group of people to test it out. Now, what does that small group of people means? I don't know if it's 3, 000 or 30, 000. Either way, they have over a hundred million users active and probably 200 million people registered. So whatever that number is, it's probably not three people, but it's probably not very large. And once they figure out all these kinks, I assume they're going to roll it out to everybody, including any user of ChatGPT that is going to change. Everything, as I mentioned before, once this can be connected to other software as the interface, nobody will want to use a keyboard and mouse anymore because you can just have a conversation, explain what you want, and the AI will perform the tasks for you in the software that you're operating. I don't see that happening this year, but I definitely see that happening in 2025. And the last small but exciting piece of news from OpenAI is they're now allowing free Chachapiti users to generate two images per day with DAL E3. So DAL E3 is the image generation, text to image generation model from OpenAI and it's been available to paid users for A very long time, I'm using it regularly, and now they're going to allow free users to use the model for, as I mentioned, two images per day. Now, why is that important? I use DALI for image generation for multiple purposes. When I need higher quality, I use other models. When I need text in images, I use other models. So I use MidJourney for the higher resolution images, and we're going to talk about an option for that. That just came out really exciting this week. And I use ideogram every time I want to create images with text in them because it's just amazing at it. So if you've never tried ideogram, go and test it out if you need any kind of graphics that involves text in it or just different styles of text that you can then copy and paste on whatever graphics that you have. But I use DALI 3 every time I need context, and it's actually the best tool for that reason. So it's the best tool when that is your need. So let me explain to specific use cases. Let's say you're developing a presentation for whatever business need. So I'm working on my presentations and brainstorming what needs to be in different slides. And researching things using chat GPT, but then at the end of this, because it has a very deep understanding of my target audience and who I am and what I'm trying to achieve and what's the goal of the presentation and what is the flow of the presentation, because I've worked on the presentation together with it, I can ask it to recommend. What kind of images should go with each slide, ask you to explain to me in text, what those might be. Then I pick the one I like, I make whatever modifications I want by explaining what modifications I want. And then I asked chat GPT to create the images for me. I do the same thing when I create some blog posts. And I do the same thing when I create social media posts. And so it's knowledge of everything that I've done for that particular goal makes it a lot easier to create images that are relevant to that. And I just find it a very Straightforward flow, different than working image by image and trying to fix it while I'm working with mid journey or ideogram. So as I mentioned, if you have the free version, first of all, I suggest paying for the full version. It's the best 20 bucks a month you've ever going to invest in any piece of software. But if you don't want to do that, you can get access to two images per day on ChatUPT free version. And the last piece of news from open AI actually doesn't come from open AI, but directly related to them comes from Elon Musk. So if you remember, Elon Musk sued open AI a while back, and he dropped the lawsuit a couple of months ago. Without explaining exactly why, but the previous lawsuit was about the fact that they're betraying their fiduciary responsibility as a non profit organization. Now, the lawsuit that he claims is significantly stronger is saying That OpenAI has breached its founding mission to develop AI for the benefit of humanity. And that Sam Altman and Greg Brockman, that I as I mentioned just took a sabbatical, manipulated must to fund while co founding OpenAI using false statements. So basically they're claiming they used fraudulent ways in order to make him fund their initial steps. So the actual lawsuit is about violation of federal racketeering laws and conspiracy to defraud Musk himself. It was pretty obvious in the previous lawsuit that he's going to lose, especially after OpenAI shared that he himself suggested to buy and basically roll OpenAI into Tesla. So his suggestion That moving open AI from a non profit organization to a for profit organization is something he personally suggested and was trying to pursue. So he dropped that lawsuit and now there's a different version of it. There's a lot of bad blood between Elon Musk and Sam Altman specifically about the role of open AI in the universe that, as I mentioned, Elon Musk was a co founder of and wrote some of the original and probably bigger checks in the beginning of the company. Now, where's that going to go? I don't know. As I mentioned, there's a lot of Turmoil going on in open AI right now, that's just add some fuel to the fire. I don't think he's going to win this lawsuit because I think it's going to be very hard for him to prove that they were planning this thing all along versus this is how it evolved. But that being said, as I mentioned, that's going to make the whole thing more interesting and I will report to you as things progress. Now, I mentioned earlier that OpenAI slashed the cost of ChatGPT to keep them more competitive in this highly competitive market. they don't live in a vacuum and Google immediately the next day slashed the prices of Gemini 1. 5 flash model. So Gemini 1. 5 is their smaller model and they just cut its cost by 80 percent starting on August 12th. So it's not yet available, but the new pricing is going to be. 0. 75 cents per million input tokens and 0. 3 dollars for every output tokens. That's going to be 50 percent cheaper than OpenAI's GPT 40 mini, which is their smaller model. That being said, from a capability perspective, GPT 4. 0 Mini is performing significantly better than Gemini 1. 5 Flash. And when I say significantly better, on most benchmarks, it's doing much better. And GPT for old mini is also performing much better on the chatbot arena. We mentioned earlier. It is currently ranked number three, just after GPT 4. 0, it's bigger brother, basically. So the GPT 4. 0 mini, they're very close on the ranking, which tells you that GPT 4. 0 mini being a smaller, much cheaper model is probably the best way to go right now if you're looking for something fast and small. And number one is still, by the way, Gemini 1. 5 pro. So the latest big model from Google. But Google's small model, Gemini 1. 5 Flash, is only ranked Number 17 on the list. So very far behind. So from a capability perspective, Gemini 1. 5 Flash is still behind GPT 4. 0 Mini, but it's going to be much cheaper than need to use. So I'm sure developers will find the right use cases for each them. But as you can see, this competition is intensifying, especially when there are more and more very capable open source model like Llama 3. 1 that was released two weeks ago, as well as other open source models, which we're going to talk about later on in this news episode. Now to put things in perspective for a minute with this whole down spiraling price. competition. There was a very interesting article by David Kahn from Sequoia and he wrote a post called the 600 billion dollar question. This is actually a follow up from an article he wrote in September of 2023 called the AI's 200 billion dollar question. And basically what he's showing is in order to get a positive returns on the huge investments that these companies are making on chips is not even talking about the rest of the infrastructure. These companies jointly need to make as of right now, 600 billion in revenue. They're not even in the ballpark. So if you remember last week, we talked about the revenue from open AI opening. I will probably make this year anything between 3. billion. Google has not shared yet what kind of new revenue it's generating from AI. But even if it's in the same ballpark and the same with Microsoft, we're talking about double digits in the teens from all these companies combined, even if you throw in Anthropic and some of the other companies, and they need to be making 600 billion instead of 12 or 16, or even 20, if you go wild with your assumptions, and so that's not a sustainable business model, but all these companies. The big ones have so much cash they're sitting on that they can bet that this will actually going to work. And I think their biggest fear is what happens if they do not bet, and this actually generate these kind of returns. So it's a very weird game that they're playing with very high stakes where they're throwing. Insane amounts of money, bigger than any company has ever thrown on a single initiative ever before, assuming that it's going to pay off. and fearing the situation where they don't make that bet and somebody else actually makes it and is able to generate that kind of revenue. Now, still about Google and interesting research came out this past week from Sonata Insights, sharing that Google search traffic has not declined despite the rise of AI search engines. So Google's search traffic has actually grown by 1. 4 percent from May 2023 to May 2024. Now that's despite the fact that perplexity that I'm a very big fan of, that is probably the most capable AI search engine right now, Has grown 42 percent from earlier this year to now, but that's still a very small number of searches compared to Google's complete dominance in search. So people has been searching Google 290 times more than people have been searching perplexity in May of 2024. So that's a very recent data that kind of shows you the trend. In addition, Google users perform about 200 searches per user per month, while perplexity average user performs about 15 searches per month. I can tell you that advanced users like me, I use perplexity now more than I use Google. I think the transition to a I based search that gives you an answer versus just a list of websites to visit and research yourself is something that is going to happen. It's going to happen across the board. Google themselves are testing it as well. I think what's going to make the biggest difference in this particular field is the recent release Of search GPT, which is the AI based quote unquote search engine that is starting to roll out from open AI. The reason I'm saying it's going to make us a much bigger impact is not necessarily because I think it's better than perplexity. I haven't had a chance to test it myself. I've seen some very positive initial reviews from people who got access to it. But the reason is perplexity does not release the amount of users they have, but let's be very loose and positive for them. Let's say they have 15 million users that are using perplexity. OpenAI has hundreds of millions of users. So if OpenAI rolls this out, that's going to be a very different kind of impact from just a user based perspective to what perplexity is doing right now. And this may. Put a dent into what Google is doing, putting more pressure on them to start changing what they're doing. What does it mean to you as a business? What it means to you as a business that this thing is happening, whether it's going to happen this quarter or this next year, doesn't really matter. If your, Business depends, or some of your revenue depends on organic traffic. And especially if most of your business depends on organic traffic, you need to start diversifying your sources, start a podcast, have a YouTube channel, set up local meetings, whatever it is, start a newsletter, anything that will give you traffic and eyeballs on what you're doing and the offering that you have that does not come directly from Google search. Now, the biggest news about Google search this week is actually not the fact that it hasn't been shrinking, but the fact that there has been an antitrust trial against Google for their dominance in the search industry, and they lost that trial. Now, this may lead to some very interesting results that are not directly related to AI, but has everything to do with what we just talked about right now, which is dominance in search. So I don't know if you know that, but Google has been making gigantic payments of billions of dollars to companies like Apple to make Google search the default search engine on all Apple devices. This will most likely be banned as a result of this latest trial. That means that will open the door for other players to become the search engine on Apple devices. This may be through other mechanisms, as we already know apple already has an agreement with chat GPT to include this in their new operating system and in their new AI offering under the Apple umbrella. Maybe they will use GPT search as their default search engine on the Apple devices and so on. This can make a very significant difference on how search is divided around the world. Another player that may jump into this is obviously Microsoft. Microsoft previously offered Apple 100%. of search ad revenue, just to become the default search engine on iPhones and iPads. And they refused because they probably got more money from Google. Now, as of right now, Google has a complete dominance. They control 95 percent of search on mobile in the U S and 84 percent on desktop with being, being a far second at 7. 8%. So nothing even in the same ballpark. What will that lead to? I don't know. Again, this will put a lot of pressure on Google from two different directions. As I mentioned, one of it is innovation of new players and now bigger players that are getting into their field. And the other is regulatory pressure that will take some of the tools that they have today that was helping them maintain their dominance. So far we talked about OpenAI and Google and both these companies has made a really big investment in a startup called Harvey, which is an AI legal startup, and they've invested a hundred million dollars together with some other really big names with a new valuation of 1. 5 billion for that company. That company was only founded in 2022 and their current valuation is 1. 5 billion. Again, showing you how much money is pouring into the AI field. What this company is doing is they've developed a chat bot that is focused on answering legal questions. It's currently achieving 86 percent level of accuracy in answering legal questions correctly. And the goal is obviously to allow people to lawyers and paralegals and law firms to do their work significantly faster and more productive. Something interesting about this firm is while it's a new company, it's attacking or trying to get to the biggest names in the world to use its tool. So their client base includes major law firms like A& Sherman and PwC, one of the big four accounting companies. So very large organizations starting using their software. So I have several thoughts on this. One is it's only achieving 86 percent accuracy. Which means it gets it wrong 14 percent of the time. The problem is that companies and individuals, whether in law firms or other places where started using similar models from other providers will become dependent on these tools and will not check its work and will still get it wrong. In this particular case, a pretty high percentage of time. I don't want my lawyers to get the information wrong 15 percent of the time. I want them to get it correct. Okay. 100 percent of the time. So 14 percent getting it wrong in a law firm is really bad. It's very different than I'm writing a blog post and it's not 100 percent perfect. So there's still issues with that technology, even when you're raising a hundred million dollars and you're raising a hundred million dollars from some of the biggest names in the world. So that's problem number one. And you all need to be aware of that when you're doing rag processes for your internal chatbots and so on. Some of the times you will get it wrong. And you need to ask yourself what's the implications of the chatbot getting specific information wrong as far as harming your company. The other thing that I wanted to mention that's very important about this particular topic is the future changes of business models. So in this particular case, it's law firms, and it's very obvious. And I shared that in several different cases in the past, but it's very obvious to me that the concept of paralegals is going to disappear from this world. It may not happen this year. It may not happen next year, but it's happening in the next few years. Why? Because these AI tools, we will be able to do. all the work that paralegals are doing, a hundred percent of it in seconds, instead of hours, days, and weeks. Now, why does that matter from a business perspective? It matters because law firms make stupid amounts of money from billing paralegal hours. And if that goes from, Oh, 20 percent of our income comes from billing paralegal hours, zero in three years, As a law firm, you need to be prepared for that and either be aware and plan accordingly or find other revenue sources to replace billable hours. Now, a similar thing is going to happen to anybody working on billable hours, including lawyers, consultants, and so on, because the actual work that you're doing will be able to be done maybe still by you, but significantly faster. So if currently all your clients together, are paying you X amount of money based on billable hours, that X may be slashed by an order of magnitude. So now you're only making 10 percent of what you made before providing the same amount of work. What does that mean to your workforce? What does it mean to your training? What does it mean to the livelihood of your company or your industry? Nobody has answers to these questions, but if you're running these kinds of companies that are billing per hour, you need to be prepared for that and start thinking about your plan on how to address it. We talked about Google's antitrust situation. The Department of Justice just announced that it's investigating complaints against NVIDIA for alleged abuse of their market dominance. So some of the customers are complaining that NVIDIA is pressuring them to buy everything from NVIDIA and not buy anything else from their competitors or else, meaning other words, we're going to prevent from you using any NVIDIA capabilities. This is obviously illegal and falls under antitrust as well. This is only in initial steps of investigation. But it's very obvious that Invisia's current control of over 90 percent of GPU market, which the entire AI industry is dependent on for training, is allowing them to do things like that and strong arm companies to do things that they may not want to do. And the fact that the DOJ is investigating that is actually good news. Now, speaking about NVIDIA, I told you that they shared some interesting things on Seagraph, which is a conference that was a week and a half ago. One of the things they shared that is actually really interesting, and I started looking into this past week, is called James. And James is a new digital human infrastructure that they're allowing customers to use. It's a highly realistic human face that can express emotions through the animation of the face. If you just Google NVIDIA James, you'll be able to watch videos of their avatars and how realistic they look, even from up close, including the movements of their eyelashes and their eyes and so on, creating eye contact, changing expressions, etc. This is going to become a part of NVIDIA's ACE technology under their NIM microsystems that they're enabling anybody to use. So there's several companies who provide these kind of tools right now. I use Haygen regularly to create videos of myself or other avatars for multiple purposes. But this is a whole different level as far as the level of fidelity that it provides and the emotional characteristic that the faces that these face expressions can create. And so it's the next generation of the same thing. And it's not a surprise, but it's definitely a big step in the direction of not being able to tell when you're talking to somebody on the screen, whether that's a real person or an AI generated entity. Going back to what impact that can have in the business world, virtual assistants, customer service agents, sales agents, game characters, like you name it, there's multiple industries are going to go through significant. Changes that will take away jobs of millions of people, because if you don't need customer service agents anymore, because these tools can be emotional, you look human and be connected to every data source, know everything and be able to change anything for the user in seconds, speak any language and never sleep and cost significantly less. Then having a customer service agent, that's very obvious where this is going. If you remember earlier this year, Klarna started implementing a chat bot that did the work of 700 full time employees. So this is just step one in this direction. And As I mentioned, the implications of this on the workforce and society are significant. And I don't think anybody is thinking about what that means to us overall. Now from my perspective, the most exciting news of this week, because I'm a geek and I like playing with AI tools and I really like creating images with AI, a new company came out of stealth. They're called black forest labs, and they just launched flux one, it's an AI text to image generator that is absolutely incredible. Now, play with it yourself. You can use it for free on their platform. You can also use it as an API. It's an open source model founded by people that have left stability AI. So some of the original founders of stability AI has founded this company, Robin Rombach, Patrick Essar and Andres Blattman, as I mentioned, all three previously working on stable diffusion now started this company. They released three different types of model Pro, Dev, and Schnell, the Pro is a closed source available via API, and you can also test it on the website. They also released a dev environment that is available and it's an open weights non commercial model that people can use for research and developing it, as well as the Schnell, which means fast in German, and it allows you to run tests similar models on a smaller scale, faster. Now the output quality is absolutely incredible. Because it's open source and a very efficient open source model, you can actually run it on your local computer and get incredible results. It's not going to run very fast, but it will generate highly realistic images that can rival. Meet journey version 6 or even 6. 1 that was just released and definitely DALI 3. Extremely exciting, very interesting. It will be very interesting to see what other companies are doing with the fact that now there's access to a top of the line text to image generation model that is open source. Now, as impressive as it is, this is just version one. So there's definitely another player to pay attention to in the image generation field. Now in parallel to that, Me Journey released Me Journey version 6. 1. It's not a major improvement and it's not a major release, but there's still improvements. Dave. Improved some aspects of image coherence, such as arms and legs and hands in people, they presumably improve the details on further away faces, which was always a problem on me journey. So if you look at images of several different people, or if one, one person from further away, it's always lacking detail. And really the only way to get good details was to create the image of the face first and then have. Mid journey zoom out and out paint around it to keep the images so they've changed that. It's not a huge difference. I assume there is a difference. I couldn't really tell if it's that good, but this release also includes a better upscaler, more details to smaller features, improve text accuracy, which is actually visible. So it's actually generating better text than it did before compositions now are better when you create them texture of different things is significantly better. The images look sharper on the smaller details. So there is noticeable changes. I don't think the upscale is that good. I still use my other open source scaler that I use when I need to upscale images. But overall, a better version from Mid Journey, and again, from anybody who's using Mid Journey to generate images, it's a nice step in the right direction. And now that there's competition, it will be interesting to see how fast the release new versions we talked about NVIDIA and their total control over the GPU world, but there's a startup that I mentioned several times before that I actually really everything that they're doing. That's called Grok. That's Grok with a Q versus Grok with a K. Grok with a K is El Musk's AI model for Twitter. Well, grok with AQ just raised$640 million at a$2.8 billion valuation led by some very big names like Cisco Investments, Samsung Catalyst and BlackRock, private equity partners. Their valuation was$1.1 billion in 2021, so they more than doubled in just three years. And what they do, their chips has a completely new architecture than GPUs. They call them LPUs, language processing units, and they do amazingly well in inference, which is basically the generation phase of these models. So these models are trained using a huge amount of data on GPUs. And then most of us are still using GPUs to generate stuff, the usage of the AI when we give it information and get information back. But this new architecture actually does it between the GPUs. 10 to 1000 times faster, depending on the use case, then the GPUs are doing it while consuming significantly less electricity. Now, Grok, in addition to just being an AI hardware platform, also has a Grok community and a Grok cloud solution where they're integrate their hardware with open source models, you can run multiple different models, and especially the latest Lama model on Grok at speeds that are unparalleled in any other platform in the AI world. and they're continuously improving Grok cloud and adding more and more functionality, as well as more and more capabilities and languages to this platform. They don't have any big known names on the leadership I've met their CTO at a conference a couple of months ago, and he's a really fascinating guy, former Alphabet top engineer. The only big no name is that they have Yan LeCun, who is Meta's chief AI scientist, as a technical advisor, which explains their close cooperation with Meta's open source models. Speaking of open source models, a few announcements on that front this week as well. So LG, the South Korean tech giant just announced that they're releasing a new open source model. They call Exxon 3. 0. They had obviously two versions before that. It's a 7. 8 billion parameter model, and it's doing very good at both Korean and English. It is not as powerful as the leading models of the West, but it's definitely giving some run for the money for the next tier models, while it's 56 percent faster than the average model out there, and a 35 percent decrease in memory, which leads to a 72 percent reduction in performance. Operational cost compared to the previous model that they had and the average of the second tier models in the industry, right now. They're planning to build a significant open source AI ecosystem in Korea, and because of their dominance in the Korean tech field, that's very likely to happen. So in addition to all the Western hemisphere companies that we talk about a lot, we talked a little bit in the past about the fact that alibaba released an open source model called Quen that is very capable and the UAE has released Falcon that is an interesting open source model. So there's growing competition around the world, mostly on the open source universe, building new and advanced models that they're going to continue to push. And speaking of open source models, it's very hard to talk about that industry without talking about Mistral. So Mistral is a French company that has been pushing very capable open source AI models for a very long time since it's beginning and they just released three open weight language models for three different purposes. I'm not going to dive into the details because it's less relevant. One of them is very interesting because it has 128,000 tokens context window, which is significantly bigger than everything they released before and is aligned with GPT 4 version of ChatGPT. For me, it's only second to none. the only larger context window, other than that, are 200, 000 from Claude and 2 million from Gemini 1. 5 Pro, which is pretty good obviously in a league of its own as of right now. The other model is also really interesting from a context window perspective. It's called CodeStrile Mamba. And it's based on the Mamba architecture instead of Transformers. Again, I don't want to dive into the details, but it's a completely different architecture than most language models are running on today. And in theory, it has an infinite context window, meaning you can use as much information in and gets as much information out in a single chat. This is a much smaller model than the really big models we're used to. It's only has 7 billion parameters, but this might be very interesting, especially in the research side to see where that goes. The other thing that they announced, which might be more interesting for people who are listening to this podcast, is that they released an agent builder capability. They actually released two versions of the agent building capabilities. One is for non technical users, so it run on a web based user interface where you can go in, select any of their models and build really sophisticated processes with AI that can do a lot more than you can do with a single prompt on a chatbot. And again, it's built for people like me and you who are non technical, but they also released a parallel version of agent API that allows developers to build agent using external platforms and connecting to this agentic capability. In general, this is the direction the world is going. There are more and more platforms that allows us today to build AI agents that will do multiple steps and complex processes. The disadvantage of doing it on a platform like Mistral is that you're limited to their models. When, if you're doing it on a third party platform, you can actually pick and choose different models for different steps of the agents that might be better suited for the simple tasks in each and every one of the steps when the overall process becomes more efficient, faster, cheaper, et cetera. And the last piece of news is just interesting to me more than anything else. And it is the fact the researchers tested an AI capability to grade K through 12 test exams of open text. And I'm quoting, LLMs can mark open text responses to short answer questions Across various by subject areas and grade levels, we found that the GPT 4 performance with minimal prompt engineering was in line with the performance of expert human raters. Going back to what I said before, I see a huge opportunity for AI to completely revolutionize education, changing it from a class based to personal based, where the teacher should become a mentor versus the one trying to teach. 20 to 35 students in the same classroom, where each and every one of them has individual needs. AI can solve that problem and really help teacher be more flexible to help students while allowing each and every one of them to progress in his or her own pace based on their own individual Teaching them in different ways, whether through videos and games or reading each person. However, he makes better progress on specific topics while relieving teachers from the need to grade work, as an example, as we just learned. Very exciting, very interesting. We will be back. that's It from a news perspective this week. There is a lot of other stuff that was not shared here that we curated during the week. If you want to get access to all of that, sign up for our newsletter. There's a link for that in the chat. If you don't know that for whatever reason, we do a live session every Thursday where you can join this podcast live with the expert that we're hosting both on zoom and on LinkedIn live and be a part of the conversation, ask questions, network with other people that are AI enthusiasts, but our business people just like you, so you can do that on our website. And again, there's a link in the show notes. You can open it right now and do that. And every Friday we do AI Friday hangouts. A really cool group of people that is a very laid back, non formal environment that happens at Friday, 1 p. m. Eastern, where we just get together and talk about what we've learned in AI in business this week, different system, different tools, answer questions and help each other out. Make progress in the AI revolution. So if you want to join that again, the link is going to be in the show notes. You can open your phone or your computer right now and click on the link and sign up to get notifications. When these happen, these are totally free and you're welcome to join us. And the last request. If you have been enjoying this podcast, if you're learning from it, please share it with people, literally open your phone right now, click on the share button and share this podcast with a few people you know, that can benefit from listening to us and the information that we're sharing. And while you're at it, when you have your phone in your hands and you're already on your podcasting app, please rate and review us on your favorite platform that helps us reach more people. And that could be your role in helping share AI literacy in the world. I would really appreciate if you do that. And until Tuesday, where we're going to have our next expert session, have an amazing weekend.

People on this episode

Head Shot

Isar Meitis

Host